On the basis of patterns of allele frequency variation in nuclear genes (Din et al., in press) it has been proposed that the house mouse M. musculus originated in the northern Indian subcontinent, from where it radiated in several directions to form the well-described peripheral subspecies (M. m. domesticus, M. m. ~~lusculu.s and M. MZ. custuneus). Here we use a mitochondrial DNA (mtDNA) phylogeny to test this hypothesis and to analyse the historical and demographic events that have accompanied this differentiation. This marker also provides a powerful means to check for genetic continuity between the central and peripheral populations. We studied restriction site polymorphism of samples from India and the Middle East as well as samples from the rest of Eurasia and northern Africa. M. m. domesticus and M. m. muscdus are both monophyletic for mtDNA and belong to the subspecies-specific mtDNA lineages that have been described previously. Average nucleotide diversity is low in M. 112. mu,r~ulu.s (0,2L5%). It is not only higher in M. m. domesticus (0.7-0.9%) but the distribution of pairwise divergence is wider, and the rate of evolution in this branch appears to be higher than in M. m. mus~u/us. The nucleotide diversity found in M. m. castaneus (0.4%) is due to the existence of two rather divergent linages with little intralineage variation. These two lineages are part of a diversified bush of the phylogenetic tree that also comprises several previously undescribed branches and includes all samples from the northern Indian subcontinent and Iran. The degree of diversity found in each of the samples from this region is high (1.2Z2.4%) although they come from small geographic areas. This agrees well with the idea that the origin of the radiation was in the northern Indian subcontinent. However, as neither haplotypcs on the M. MI. r1on~c.vtia1.v nor on the M. 171. ~~~cr.sc~ulus branches were found in this region, there appear to be important phylogeographic discontinuities between this central region and these periphcrial subspecies. On the basis of the present results and the nuclear data (Din et al., in press), we propose that M. n~u.cc~u1zr.s originated in the north of the Indian subcontinent. Our calibration of the evolutionary rate of mtDNA in mice suggests that the mouse settlement in this region could be as old as 900 000 years. Possibly from there, a first radiation could have reached the Middle East and the Caspian Sea, where the M. 1~. ~lot~~~stic~~~.s and M. tw. t~~u.s~~du.s lineages, respectively, would have started to difrcrentiate a few hundred thousand years ago, and from where they could have coloniscd the peripheral part of their ranges only recently. M. MI. c~ostmcus appears from its mtDNA to be a recent offshoot of the northern Indian population. This multiple and gradual radiation ultimately led to recent peripheral secondary contacts, such as the well-known European hybrid zone.

The extent to which alleles can disperse across a hybrid zone depends on the selection they are subjected to in the hybrid genetic background or, for those that are selectively neutral, on their ability to escape from the unfavourable environment by recombination. Three markers spanning a 45 CM segment in the center of the X chromosome were used to investigate the degree to which selection against X chromosome linked genes helps to maintain the barrier to gene flow in the hybrid zone between A4us musculus domesticus and M. m. musculus in Denmark.The introgression of all the sex chromosome specific markers was more limited than that of the autosomal enzymes (Idh I, Amy, Gpdl, Pgml, Esl, Es2. Mpi, Npl, EslO, Sod]) and the mitochondriul DNA. The cline for DXPus2, which is in the center of the X chromosome, is extremely steep and shows that certain genes located in this region are strongly selected against in the hybrid background.The clines of the other two X-linked markers, Hprt and DXPusl, and of the Y chromosome are not as abrupt and all three have similar asymmetric introgression patterns. Although the musculus variants appear to behave in much the same way as those of the autosomal genes, the domesticus variants do not introgress. The results show that X-linked and to a lesser extent Y-linked genes are more strongly selected against in the hybrid genome than the mitochondrial genome or the different autosomal loci. This suggests that co-adapted gene systems involving the sex chromosomes may play an important role in the hybrid breakdown between the two subspecies.

We analysed the patterns of allele frequency change for ten diagnostic autosomal allozyme loci in the hybrid zone between the house mouse subspecies Mus musculus domesticus and M. m. musculus in central Jutland. After determining the general orientation of the clines of allele frequencies, we analysed the cline shapes along the direction of maximum gradient. Eight of the ten clines are best described by steep central steps with coincident positions and an average width of 8.9 km (support limits 7.6-12.4) flanked by tails of introgression, indicating the existence of a barrier to gene flow and only weak selection on the loci studied. We derived estimates of migration from linkage disequilibrium in the centre of the zone, and by applying isolation by distance methods to microsatellite data from some of these populations. These give concordant estimates of s = 0.5-0.8 km generation -. The barrier to gene flow is of the order of 20 km (support limits 14-28), and could be explained by selection of a few per cent at 43-120 underdominant loci that reduces the mean fitness in the central populations to 0.45. Some of the clines appear symmetrical, whereas others are strongly asymmetrical, and two loci appear to have escaped the central barrier to gene flow, reflecting the differential action of selection on different parts of the genome. Asymmetry is always in the direction of more introgression into musculus , indicating either a general progression of domesticus into the musculus territory, possibly mediated by differential behaviour, or past movement of the hybrid zone in the opposite direction, impeded by potential geographical barriers to migration in domesticus territory.Barton NH, Gale KS. 1993. Genetic analysis of hybrid zones.In: Harrison RG, ed. Hybrid zones and the evolutionary process. Oxford: Oxford University Press, 13-45. Barton NH, Hewitt GM. 1985. Analysis of hybrid zones.